Here’s a bit of a puzzle about memory. An ordinary view, in the language of modularity, is that the memory box contains propositions, and when one explicitly recalls something, one accesses such information from one’s memory box (though this claim is not meant to suggest that there aren’t other things going on in memory that are more image-based).
But now consider the question of the justification of a particular kind of belief. Suppose several years ago you visited me at my former department and were impressed with the outstanding live oak tree outside our building. You perceived the tree being outside the building, and came to believe that there is an outstanding specimen of a live oak tree there.
Now three years later you still believe that there is an outstanding specimen of a live oak tree there (if you’re worried that the span of time is too long for justification to be present, we can shorten it to allay that worry). What is your present belief based on?
Is it based on the initial perception? That looks quite a bit like action at a temporal distance.
Is it based on memory? If so, how? It’s not as if you are in the seeming state of remembering the tree. You simply still hold the belief you formed quite a bit ago. You could make memory operative by bringing to mind the visit and the experience, but you needn’t do that to continue to belief that there is a three at that location.
The temptation is to say that the belief itself, and not (just?) the content of the belief, is stored in the memory box, and that’s how memory is presently responsible for the belief and how that belief is based on memory. But that is a rather anti-modularity idea.